File size: 839 Bytes
0301a69
 
68facde
 
 
0301a69
 
 
 
 
 
 
68facde
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
---
title: Swinunetr Dicom Video
emoji: πŸ“–πŸŽ¬
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 3.0.24
app_file: app.py
pinned: false
license: apache-2.0
---

This repository contains the code for UNETR: Transformers for 3D Medical Image Segmentation. UNETR is the first 3D segmentation network that uses a pure vision transformer as its encoder without relying on CNNs for feature extraction. The code presents a volumetric (3D) multi-organ segmentation application using the BTCV challenge dataset.

Check out the Beyond the Cranial Vault source Swin-UNET models [here](https://huggingface.co/darragh/swinunetr-btcv-small). Also in the link, you can see links to the original BTCV winning solution.

This is a small demo on a subset of the test data for the [BTCV competition](https://zenodo.org/record/1169361#.YtGvn-xKhb8).